CN111772627B - Online fetal electrocardiosignal extraction device and method - Google Patents

Abstract

The invention discloses an online fetal electrocardiographic signal extraction system and a method thereof. The method includes: firstly, collecting abdominal signals of pregnant women; then dividing the abdominal signals of pregnant women to obtain current window signals, and finding the R wave position of maternal electrocardiographic signals; and then According to the R wave position of the maternal ECG signal, the current window signal is segmented to obtain several signal segments; then, according to the signal segments in the current window signal, the maternal ECG signal template belonging to the current window signal is reconstructed; and then according to the current window signal The maternal ECG signal template of the window signal is calculated to obtain the fetal ECG component in the next window signal; finally, the above steps are repeated until the fetal ECG components of the remaining signal windows except the first signal window are obtained. The method includes a data preprocessing method, maternal electrocardiographic component reconstruction and fetal electrocardiographic component extraction. Through the technical solution, the fetal electrocardiogram signal can be accurately extracted online.

Description

An online fetal ECG signal extraction device and method

technical field

The invention belongs to the technical field of biomedical signal processing, and in particular relates to an online fetal electrocardiographic signal extraction device and method.

Background technique

At present, there are three main types of clinical monitoring of fetal growth and development: fetal heart sound monitoring, cardiac monitoring and ECG monitoring. Their functions are to detect abnormalities in the process of fetal growth and development early by monitoring fetal heart activity during pregnancy. Fetal ECG signal is the source signal of fetal heart activity. Compared with heart sound and heartbeat signal, it is the physiological signal that can best reflect the whole picture of fetal heart signal. The online extraction of fetal ECG signals can reflect the biophysical activity of the fetus in real time, which is of great significance.

The methods for obtaining fetal heart rate signals mainly include scalp electrode method and abdominal electrode method. The scalp electrode method is an invasive method that directly contacts the fetal scalp and can obtain clear fetal ECG signals. This method is traumatic and prone to infection of the pregnant woman and fetus, so it is often only used during childbirth. The abdominal electrode method is a non-invasive method or an indirect method to obtain multiple sets of mixed signals of maternal and fetal ECG signals through electrodes at different positions in the pregnant woman's abdomen. Due to interference from the body surface of pregnant women, poor conductivity of human tissue, and complex external environmental interference, the signal obtained by the abdominal click method is a mixture of pregnant women's ECG signals, fetal ECG signal noise interference, and random signals. Therefore, the abdominal electrode method needs to be combined with an excellent extraction algorithm to extract clear fetal ECG signals.

At present, there are some extraction methods of fetal ECG signals, such as adaptive filtering method, methods based on singular value decomposition, principal component analysis and independent component analysis, etc., but there are still shortcomings that most of the methods work in offline mode and cannot be processed in a timely manner. Reflect the time-varying characteristics of the fetal ECG signal.

Contents of the invention

The purpose of the present invention is to overcome the shortcomings of the above-mentioned existing methods, and propose an online fetal ECG signal extraction system and method, which can well remove the mother's ECG signal and extract clear fetal ECG signals.

On the one hand, the present invention provides a kind of fetal electrocardiogram signal online extraction method;

Include the following steps:

Collect abdominal signals of pregnant women;

Segment the abdominal signal of the pregnant woman to obtain the current window signal, and find the R wave position of the maternal ECG signal;

According to the R-wave position of the maternal ECG signal, the current window signal is segmented to obtain several signal segments;

According to the signal segment in the current window signal, reconstruct the maternal ECG signal template belonging to the current window signal;

According to the maternal ECG signal template of the current window signal, the fetal ECG component in the next window signal is calculated.

In one of the possible embodiments, when segmenting the abdominal signal of a pregnant woman to obtain the current window signal, the first m window signals are the first window signal, and the current window signal after the mth one is the second window signal; wherein, the The length of the first window signal is T1, the length of the second window signal is T2, and the T1>T2;

In a certain embodiment, the length of the first window signal is 20-40s, preferably 30s;

In a certain embodiment, the length of the second window signal is 5-15s, preferably 10s;

In a certain embodiment, m is greater than or equal to 2; more preferably, m=2;

In a certain embodiment, the fetal ECG components belonging to the first window signal are not calculated or the fetal ECG components belonging to the first window signal are removed.

In one of the possible embodiments, when looking for the R-wave position of the maternal ECG signal, it includes: extracting the peak in the window signal through the first-order difference sequence of the current window signal, and using the peak amplitude as the clustering variable , using a clustering algorithm to extract the position of the R-wave peak of the mother's ECG signal;

In a certain embodiment, when using a clustering algorithm to extract the position of the R wave peak of the mother's electrocardiogram, the following steps are included:

Determine the number of separate clusters and select the cluster center of each cluster; the number of clusters is 2;

Calculate the Euclidean distance from the R wave peak to each cluster center;

According to the Euclidean distance from each R-wave peak to each cluster center, obtain the cluster center corresponding to each R-wave peak;

Calculate the sum of the Euclidean distances from the R-wave peaks to their corresponding cluster centers;

Then select another new cluster center;

Compare the sum of the Euclidean distances from the R wave peak to the new corresponding cluster center and the old corresponding cluster center, and select the smaller one;

Finally, the cluster center with the closest sum of the Euclidean distances from the R wave peak to the respective corresponding cluster centers is obtained, and then the cluster center with the largest x value and its corresponding R wave peak are selected as the maternal ECG signal R wave peak;

In a certain embodiment, the distance formula is a Euclidean distance formula, and the Euclidean distance formula is Among them, y is the peak value of each spike, x is the peak value of the cluster center, and n is the number of spikes.

In one of the possible embodiments, after segmenting the current window signal to obtain several signal segments according to the R wave position of the maternal ECG signal, it also includes filtering the signal segments according to a preset threshold;

In a certain embodiment, the signal segments satisfying "b ≤ T _RR ≤ a" are selected, wherein the n signal segments that meet the requirements are arranged according to the length of time, and the arranged signal segments are Xi, i=1～n , and T _RR is the time length of the signal segment, a is the time length of X(n–([n/10] rounded)), b is the time length of X([n/10] rounded).

In one of the possible embodiments, when reconstructing the maternal ECG template belonging to the current window signal according to the signal segments in the current window signal, the following steps are included: according to the time length of each signal segment in the current window signal Obtain the standard time length; extend or compress the time length of each signal segment to the standard time length; obtain the maternal ECG signal template according to the extended or compressed signal segment;

In a possible embodiment, after averaging the extended or compressed signal segments at the same time point, the average value at all time points is used to form the maternal ECG signal template. (All the extended or compressed signal segments are averaged at the same time point to obtain the value of the maternal ECG signal template at one time point, and then combine the values at all time points to obtain the maternal ECG signal template.)

In one of the possible embodiments, when the standard time length is obtained according to the time length of each signal segment in the current window signal, the signal segments are arranged according to the time length, and the length of the signal segment located in the middle is selected as the standard time length;

In a certain embodiment, when the number of signal segments is an odd number, the order is taken as the first The time length of a signal segment is the standard time length; when the number of signal segments is an even number, the sorting is the first /> The time length of signal segments is the standard time length; where, n is the number of signal segments in the current window signal.

In a possible embodiment, when calculating the fetal ECG component in the next window signal according to the maternal ECG signal template of the current window signal, the following steps are included:

Composing the parent ECG analog signal belonging to the current window signal through the maternal ECG signal template of the current window signal;

Subtract the maternal ECG analog signal of the current window signal from the next window signal to obtain the fetal ECG component;

In a certain embodiment, when the maternal ECG signal template of the current window signal is used to form the maternal ECG analog signal belonging to the current window signal, it includes: The maternal ECG analog signal with the same time length of the signal;

In a certain embodiment, a maternal ECG signal template is added before the maternal ECG analog signal;

In a certain embodiment, a template of the maternal ECG signal located at the end of the maternal ECG analog signal is completed.

In a certain embodiment, the R peak of the first parent ECG signal of the next window signal whose position of the R peak of the maternal ECG signal has been found is aligned with the second of the parent ECG analog signal of the current window signal. The location of the R peak.

In a certain embodiment, the R peak of the first maternal ECG signal in the next window signal of the position of the maternal ECG R peak that has been found is aligned with the maternal ECG analog signal (corresponding to the window signal). Before the position of the second R peak, including: looking for the R peak position of the maternal ECG signal of the next window signal; that is, calculating the R peak position of the maternal ECG signal of the next window signal through the first-order difference and clustering algorithm .

In one of the possible embodiments, after the abdominal signal of the pregnant woman is collected, the abdominal signal of the pregnant woman is preprocessed, or after the abdominal signal of the pregnant woman is divided to obtain the window signal, the abdominal signal of the pregnant woman is preprocessed; preferably, the preprocessing The method comprises processing baseline, smoothing signal curve and/or filtering clutter; more preferably, wavelet transform is used for filtering and smoothing.

On the other hand, a fetal ECG signal extraction system is provided, including:

The collection module is used to collect abdominal signals of pregnant women;

The window signal segmentation module is used to segment the abdominal signal of the pregnant woman to obtain the current window signal, and find the R wave position of the maternal ECG signal;

The signal segment segmentation module is used for segmenting the current window signal to obtain several signal segments according to the R wave position of the mother's ECG signal;

Maternal electrocardiographic signal template construction module, used for reconstructing the maternal electrocardiographic signal template belonging to the present window signal according to the signal segment in the present window signal;

The fetal ECG component calculation module is used to calculate the fetal ECG component in the next window signal according to the maternal ECG signal template of the current window signal.

In one of the possible embodiments, the window signal segmentation module is specifically used to obtain the current window signal by segmenting the abdominal signal of a pregnant woman, wherein the first m current window signals are the first window signal, and the mth following current window signals are is the second window signal; the length of the first window signal is T1, the length of the second window signal is T2, and the T1>T2.

In one of the embodiments, the window signal segmentation module is specifically used to extract the peak in the window signal through the first-order difference sequence of the current window signal, use the peak amplitude as the clustering variable, and use the clustering algorithm to extract the peak The position of the R wave peak of the ECG signal;

In one of the embodiments, the window signal segmentation module is used to determine the number of separate clusters and select the cluster center of each cluster; calculate the Euclidean distance from the R wave peak to each cluster center;

According to the Euclidean distance from each R-wave peak to each cluster center, obtain the cluster center corresponding to each R-wave peak;

Calculate the sum of the Euclidean distances from the R-wave peaks to their corresponding cluster centers;

Then select another new cluster center;

Compare the sum of the Euclidean distances from the R wave peak to the new corresponding cluster center and the old corresponding cluster center, and select the smaller one;

Finally, the peak to the cluster center with the closest distance to the sum is obtained, and then the cluster center with the largest x value and its corresponding R wave peak are selected to be defined as the R wave peak of the mother's ECG signal;

In one of the embodiments, the distance formula is a Euclidean distance formula, and the Euclidean distance formula is Among them, y is the peak value of each spike, x is the peak value of the cluster center, and n is the number of spikes.

The fetal ECG signal extraction system also includes a screening module, which is used to filter the signal segment according to a preset threshold after segmenting the current window signal to obtain several signal segments according to the R wave position of the maternal ECG signal;

In one of the possible embodiments, the screening module is used to select the signal segments satisfying "b≤T _RR ≤a", wherein, the n signal segments satisfying "b≤T _RR ≤a" are performed according to the length of time Arrangement, the arranged time period is Xi, i=1～n, and T _RR is the time length of the signal segment, a is the time length of X(n–([n/10] rounded)), b is X( [n/10] rounded) the length of time.

In one of the possible embodiments, the maternal ECG template building module is used to obtain the standard time length according to the time length of each signal segment in the current window signal; extend or compress the time length of each signal segment to the standard time Length; Obtain the maternal ECG signal template according to the extended or compressed signal segment;

In a certain embodiment, after averaging the extended or compressed signal segments at the same time point, the average value at all time points is used to form a maternal ECG signal template.

In one of the possible embodiments, when the standard time length is obtained according to the time length of each signal segment in the current window signal, the maternal ECG signal template construction module is specifically used to arrange the signal segments according to the time length, select The length of the signal segment in the middle is the standard time length;

In a certain embodiment, when the number of signal segments is an odd number, the order is taken as the first The time length of a signal segment is the standard time length; when the number of signal segments is an even number, the sorting is the first /> The time length of signal segments is the standard time length; where, n is the number of signal segments in the current window signal.

In one of the possible embodiments, the fetal ECG component calculation module composes the maternal ECG analog signal belonging to the current window signal through the maternal ECG signal template of the current window signal; subtracts the maternal ECG signal of the current window signal from the next window signal ECG analog signal to obtain the fetal ECG component (belonging to the next window signal);

In one of the possible embodiments, when the maternal ECG analog signal belonging to the current window signal is composed of the maternal ECG signal template of the current window signal, the fetal ECG component calculation module is specifically configured to use the maternal ECG signal template of the current window signal The electrocardiographic signal template is stitched together to form a maternal electrocardiographic analog signal that is consistent with the time length of the next window signal;

In a certain embodiment, a maternal ECG signal template is added before the maternal ECG analog signal;

In a certain embodiment, a maternal ECG signal template located at the end of the maternal ECG analog signal is completed;

In a certain embodiment, the fetal ECG component calculation module is also specifically configured to align the R peak of the first maternal ECG signal of the next window signal with the position of the R peak of the maternal ECG signal with the current window signal The location of the second R-peak of the maternal ECG analog signal.

In one possible embodiment, the fetal ECG signal extraction system further includes a preprocessing module, which is used to preprocess the abdominal signal of the pregnant woman after collecting the abdominal signal of the pregnant woman, or divide the abdominal signal of the pregnant woman to obtain the current window signal, Preprocessing the abdominal signal of pregnant women;

In a certain embodiment, the method of preprocessing includes processing baseline, smoothing signal curve and/or filtering clutter;

In one embodiment, wavelet transform is used for filtering and smoothing.

Compared with the prior art, the beneficial effects of the above scheme:

1. The present invention uses the current window signal of the next window signal as a template, so when the fetal ECG signal is extracted from the next window signal, there is no need to calculate the maternal ECG signal of the next window signal, which greatly improves the real-time monitoring efficiency.

2. Provide a method for extracting fetal electrocardiogram signals, which does not involve a large number of matrix operations and iterative multiplication operations, and the complexity is greatly reduced.

3. Moreover, the time length of each signal window is relatively short, greatly reducing the amount of calculation, and effectively solving the delay problem of online fetal ECG signal extraction.

4. The method for reconstructing maternal ECG signals of the present invention can effectively obtain more accurate maternal ECG signals.

Description of drawings

Fig. 1 is the step flow chart of fetal electrocardiogram signal online extraction method of the present invention;

Fig. 2 is the specific detailed step flow chart of fetal electrocardiogram signal online extraction system method of the present invention;

Fig. 3 is parent R wave crest or trough mark schematic diagram in the method of the present invention;

Fig. 4 is the schematic diagram of parent template in the method of the present invention;

Fig. 5 is a schematic diagram of the maternal ECG signal reconstructed in the method of the present invention;

Fig. 6 is the schematic diagram when the mother's ECG signal reconstructed in the method of the present invention is aligned with the next window signal;

Fig. 7 is the schematic diagram of preliminary estimation of fetal electrocardiogram in the method of the present invention;

Fig. 8 is a schematic diagram of a clear fetal ECG signal extracted in the method of the present invention;

FIG. 9 is a schematic structural diagram of a fetal ECG signal extraction system.

Detailed ways

In order to further explain the technical means and effects adopted by the present invention in order to achieve the intended purpose of the invention, the specific implementation, structure, features and effects of the present invention will be described in detail below in conjunction with the accompanying drawings and preferred embodiments.

Example 1

A method for extracting fetal electrocardiogram signals;

As shown in Figure 1, the following steps are included:

Step S1: collecting the abdominal signal of the pregnant woman; it should be understood that the abdominal signal of the pregnant woman refers to the abdominal mixed signal including the abdominal mixed signal of the maternal ECG signal and the fetal ECG signal. At the same time, it should also be understood that the abdominal signal acquisition of the pregnant woman may be single-channel acquisition or multi-channel acquisition.

Step S2: Segment the abdominal signal of the pregnant woman while collecting the abdominal signal of the pregnant woman to obtain the current window signal, and find the R wave position of the maternal ECG signal, as shown in Figure 3; it should be understood that in this embodiment, the abdominal signal of the pregnant woman can be collected at the same time At the same time, the window signal is segmented to realize the real-time extraction of the fetal ECG signal. At the same time, it should be understood that this technical solution is to determine the R wave position of the maternal ECG signal and complete the subsequent steps by finding the peak position of the R wave of the maternal ECG signal, but it can also be determined by finding the trough of the R wave of the maternal ECG signal or any R wave. A location to achieve.

Step S3: According to the position of the R wave of the maternal ECG signal, the current window signal is segmented to obtain several signal segments; that is, the signal segment is divided with the position of the R wave as the breakpoint.

Step S4: According to the signal segment in the current window signal, reconstruct the maternal ECG signal template belonging to the current window signal;

Step S5: Calculate and obtain the fetal ECG component in the next window signal according to the maternal ECG signal template of the current window signal.

It should be understood that the processing method of the next window signal in the above technical solution is the same as that of the current window signal, and steps S1 to S5 need to be completed. That is, each window signal is handled in the same way.

In one of the possible embodiments, when the pregnant woman's abdominal signal is divided to obtain the current window signal, the first m current window signals are the first window signal, and the mth subsequent current window signal is the second window signal; wherein, the The length of the first window signal is T1, the length of the second window signal is T2, and the T1>T2;

Preferably, the length of the first window signal is 30s;

Preferably, the length of the second window signal is 10s;

Preferably, m=2;

Preferably, when the above steps are repeated until the fetal electrocardiographic components of the remaining signal windows except the first signal window are obtained, the fetal electrocardiographic components belonging to the signal of the first window are not calculated. Preferably, the R wave positions of all remaining first window signals except the last first window signal are not searched for. That is to say, in this embodiment, the first 60s of the collected pregnant woman’s abdominal signal is composed of two first window signals, and the first first window signal (i.e., the first 30s) is not processed after being segmented (because there may be some Murmur, so the initial fetal signal cannot be calculated based on the initial maternal abdominal signal). The second first window signal (the last 30 s) can be used to calculate the maternal ECG template and serve as a basis for the subsequent fetal ECG components. The reasons for taking two first windows and setting the signal length of the first window to 30s are: firstly, in the absence of an initial template, it is necessary to use enough segmented data segments for averaging. There are about 35 signal segments within the 30s window, and a relatively accurate initial maternal ECG signal template (ie, the first maternal ECG signal template) can be obtained by averaging enough signal segments. Moreover, when obtaining the initial maternal ECG template, if the window signal of 10s is used immediately, the template may not be accurate enough. And adopting the first signal window of 30s can guarantee the accuracy of the data.

In one of the possible embodiments, when looking for the R-wave position of the maternal ECG signal, all peaks in the window signal are extracted through the first-order difference sequence of the current window signal, and all peak amplitudes are used as clustering variable, using a clustering algorithm to extract the position of the peak of the R wave of the mother's ECG signal; in order to better illustrate the way of finding the peak of the first-order difference sequence in this embodiment, the first-order difference sequence: record the window signal as y _i , i=1 ,...n, and then make a first-order difference, m _i =y _i -y _i-1 , i=2,...n can get the peak value of the window signal by transforming the positive and negative values of the differential data, and thus the peak amplitude can also be obtained value. The example is as follows: y=[3,4,5,4,2,8,9,2,3,7,9,12,8,6], m=[0,1,1,- 1,-2,6,1,-7,1,4,2,3,-4,-2], it can be seen from m: the peak value is: 5, 2, 9, 2, 12. The amplitude is the phase Absolute value of difference between adjacent peak values: 3, 7, 7, 10.

In one of the embodiments, when using the clustering algorithm to extract the position of the R wave peak of the mother's electrocardiogram, the following steps are included:

N1. Determine the number of separate clusters, and select the cluster center of each cluster; preferably, the number of clusters is 2;

N2. Calculate the Euclidean distance from all R wave peaks to each cluster center; preferably, the distance formula is a Euclidean distance formula, and the Euclidean distance formula is Among them, y is the peak value of each peak, and x is the peak value of the cluster center; the one-dimensional Euclidean distance formula is adopted, which is simple to calculate and improves the operation efficiency.

N3. Obtain the cluster center corresponding to each R wave peak according to the Euclidean distance from each R wave peak to each cluster center;

N4. Calculate the sum of the Euclidean distances from all R-wave peaks to their corresponding cluster centers;

N5. Select another new cluster center (combination);

N6. Compare the sum of the Euclidean distances from all R-wave peaks to the new corresponding cluster center and the old corresponding cluster center, and select the smaller one; that is, when there are two or more cluster centers, calculate When summing the distances, only the distance between the cluster centers that are closer to the R-wave peak and the R-wave peak is considered, and the distances between other cluster centers and the R-wave peak are ignored.

N7, return to step N5, until all R-wave peak combinations are selected as cluster centers, and finally get the cluster center with the closest sum of the Euclidean distances from all R-wave peaks to their corresponding cluster centers, and then select x The cluster center with the largest value (that is, the cluster center with the highest peak value) and its corresponding R wave peak were defined as the R wave peak of the mother's ECG signal. It should be understood that in addition to using the above-mentioned clustering algorithm to calculate R-wave peaks, other clustering algorithms are also possible.

In one of the possible embodiments, after segmenting the current window signal to obtain several signal segments according to the R wave position of the maternal ECG signal, more precisely, between step S3 and step S4, further includes , filter signal segments according to preset thresholds;

In one embodiment, the signal segments satisfying "b≤T _RR ≤a" are selected, wherein the n signal segments satisfying "b≤T _RR ≤a" are arranged according to the length of time from short to long, and the arranged The signal segment is Xi, i=1~n, that is, the time length of X1 is the shortest, and the time length of Xn is the longest, and T _RR is the time length of the signal segment, and a is the signal segment X(n–([n/10] takes whole)), b is the time length of the signal segment X ([n/10] rounded).

In one of the possible embodiments, when reconstructing the maternal ECG template belonging to the current window signal according to the signal segment in the current window signal, the following steps are included:

S41. Obtain a standard time length according to the time length of each signal segment in the current window signal;

S42. Extend or compress the time length of each signal segment to a standard time length;

S43. According to the extended or compressed signal segment, the maternal ECG signal template is finally obtained, as shown in FIG. 4 ;

Preferably, after averaging the extended or compressed signal segments at the same time point, the average values belonging to different time points are used to form the maternal ECG signal template. (All extended or compressed signal segments in a current window signal are averaged at the same time point, and after obtaining the value of the maternal ECG template at a time point, combine the values at all time points to obtain the maternal heart rate electrical signal template.)

In one of the possible embodiments, when the standard time length is obtained according to the time length of each signal segment in the current window signal, the signal segments are arranged according to the time length, and the time length of the signal segment in the middle is selected as the standard length of time. It should be understood that by selecting the time length of the middlemost signal segment as the standard time length, it is possible to effectively avoid deviations in some data in the early or late stages of the signal segment.

In one embodiment, when the number of signal segments is an odd number, the order is taken as the first The time length of a signal segment is the standard time length; when the number of signal segments is an even number, the sorting is the first /> The time length of signal segments is the standard time length; where, n is the number of signal segments in the current window signal.

In a possible embodiment, when calculating the fetal ECG component in the next window signal according to the maternal ECG template of the current window signal, as shown in Figure 2, the following steps are included:

S51, form the maternal ECG analog signal belonging to the current window signal through the maternal ECG signal template of the current window signal, as shown in Figure 5;

S52. Subtract the maternal ECG analog signal from the next window signal to obtain the fetal ECG component; that is, when calculating the fetal ECG component of the Yth window signal, it is necessary to calculate the maternal ECG belonging to the Y-1th window signal Signal template.

In one embodiment, when the mother's ECG signal template of the current window signal is used to form the parent's ECG analog signal belonging to the current window signal, it includes: The maternal ECG analog signal with the same time length of the signal;

In one embodiment, it also includes adding a maternal ECG signal template before the maternal ECG analog signal;

In an embodiment, the method further includes complementing a maternal ECG signal template located at the end of the maternal ECG analog signal. It should be understood that the constructed maternal ECG analog signal is not necessarily a complete maternal ECG signal template for the maternal ECG analog signal at the 10th second, so the maternal ECG signal template at the 10th second signal needs to be completed.

The above method ensures that the obtained maternal ECG analog signal must be greater than 10s by completing or adding templates, and ensures that when the next window signal is subtracted from the maternal ECG analog signal obtained through the current window signal to obtain the fetal ECG component, There will be no problems due to short maternal ECG analog signals.

In one embodiment, it also includes that when the next window signal is subtracted from the maternal ECG analog signal obtained by the current window signal to obtain the fetal ECG component of the next window signal, the maternal ECG that has been found The position of the R peak of the first maternal ECG signal in the position of the signal R peak is aligned with the position of the second R peak of the maternal ECG analog signal (corresponding to the current window signal), as shown in FIG. 6 . For example, when calculating the fetal ECG signal of the Yth window signal, use the R peak position of the first maternal ECG signal of the Yth window signal AECG processed in step S2 to align the constructed maternal ECG analog signal (Y-1th window signal) at the second R peak position, and then cancel the two signals to obtain the fetal ECG signal belonging to the Yth window signal, as shown in FIG. 7 .

In one of the embodiments, it also includes processing the fetal ECG signal to smooth curves and/or filter clutter, and more preferably, to perform filtering and smoothing processing in the form of wavelet transform, and finally obtain the first one as shown in Figure 8. Fetal ECG signal diagram of Y window signals. It should be understood that the R wave peak of the first maternal ECG signal in the next window signal (the Yth window signal) of the position of the maternal ECG R peak that has been found is aligned with the maternal ECG analog signal (corresponding to the current window signal, that is, the position of the second R peak of the Y-1th window signal), it is necessary to find the R peak position of the parent ECG signal of the next window signal, and the above R peak position can be obtained by first-order difference and clustering Algorithm derived.

In one of the possible embodiments, after the abdominal signal of the pregnant woman is collected, preprocessing is performed on the abdominal signal of the pregnant woman, or after the abdominal signal of the pregnant woman is divided to obtain the current window signal, the abdominal signal of the pregnant woman is preprocessed; preferably, the preprocessing The processing method includes processing baseline, smoothing signal curve and/or filtering clutter; more preferably, wavelet transform is used for filtering and smoothing.

Since the above method uses a relatively simple Euclidean distance formula for clustering calculation, and the time length of the signal segment is short (there will not be many peaks in the signal segment), the complexity of the entire calculation is low, and the computer can Get results faster and improve real-time monitoring efficiency.

Example 2

A kind of fetal electrocardiogram signal online extraction system, as shown in Figure 9, comprises:

The collection module is used to collect abdominal signals of pregnant women;

The window signal segmentation module is used to segment the abdominal signal of the pregnant woman to obtain the current window signal, and find the R wave position of the maternal ECG signal;

The signal segment segmentation module is used for segmenting the current window signal to obtain several signal segments according to the R wave position of the mother's ECG signal;

Maternal electrocardiographic signal template construction module, used for reconstructing the maternal electrocardiographic signal template belonging to the present window signal according to the signal segment in the present window signal;

The fetal ECG component calculation module is used to calculate the fetal ECG component in the next window signal according to the maternal ECG signal template of the current window signal.

In one of the possible embodiments, the window signal segmentation module is specifically used to obtain the current window signal by segmenting the abdominal signal of a pregnant woman, wherein the first m current window signals are the first window signal, and the mth following current window signals are is the second window signal; the length of the first window signal is T1, the length of the second window signal is T2, and the T1>T2.

In one of the embodiments, the window signal segmentation module is specifically used to extract all peaks in the window signal through the first-order difference sequence of the current window signal, and use the clustering algorithm with all peak amplitudes as clustering variables Extract the position of the R wave peak of the mother's ECG signal;

In one of the embodiments, the window signal segmentation module is used to determine the number of separate clusters and select the cluster center of each cluster; calculate the Euclidean distance from the R wave peak to each cluster center;

According to the Euclidean distance from each R wave peak to each cluster center, obtain the cluster center corresponding to each R wave peak; the number of cluster centers can be 2.

Calculate the sum of the Euclidean distances from the R-wave peaks to their corresponding cluster centers;

Then select another new cluster center (combination);

Compare the sum of the Euclidean distances from the R wave peak to the new corresponding cluster center and the old corresponding cluster center, and select the smaller one;

Then select another new cluster center, compare the sum of the Euclidean distances from the R wave peak to the new corresponding cluster center and the old corresponding cluster center, select the smaller one, until all R wave peaks (combination) have been selected as the cluster center, and finally get the closest cluster center of the sum of the Euclidean distances from all R-wave peaks to their corresponding cluster centers, and then select the cluster center with the largest x value (that is, the cluster with the highest peak Center) and its corresponding R wave peak are defined as the R wave peak of the mother's ECG signal;

In one of the embodiments, the distance formula is a Euclidean distance formula, and the Euclidean distance formula is Among them, y is the peak value of each spike, x is the peak value of the cluster center, and n is the number of spikes.

The fetal ECG signal extraction system also includes a screening module, which is used to filter the signal segment according to a preset threshold after segmenting the current window signal to obtain several signal segments according to the R wave position of the maternal ECG signal;

In one of the possible embodiments, the screening module is used to select the time period meeting "b≤T _RR ≤a", wherein, the n time periods meeting "b≤T _RR ≤a" are sorted according to the length of time Arrangement, the arranged time period is Xi, i=1～n, and T _RR is the time length of the signal segment, a is the time length of X(n–([n/10] rounded)), b is X( [n/10] rounded) the length of time.

In one of the possible embodiments, the maternal ECG template building module is used to obtain the standard time length according to the time length of each signal segment in the current window signal; extend or compress the time length of each signal segment to the standard time Length; Obtain the maternal ECG signal template according to the extended or compressed signal segment;

In a certain embodiment, after averaging the extended or compressed signal segments at the same time point, the average values at different time points are used to form the template of the maternal ECG signal.

In one of the possible embodiments, when the standard time length is obtained according to the time length of each signal segment in the current window signal, the maternal ECG signal template construction module is specifically used to arrange the signal segments according to the time length, select The time period in the middle is the standard time period;

In a certain embodiment, when the number of signal segments is an odd number, the order is taken as the first The time length of a signal segment is the standard time length; when the number of signal segments is an even number, the sorting is the first /> The time length of signal segments is the standard time length; where, n is the number of signal segments in the current window signal.

In one of the possible embodiments, the fetal ECG component calculation module composes the maternal ECG analog signal belonging to the current window signal through the maternal ECG signal template of the current window signal; subtracts the maternal ECG signal of the current window signal from the next window signal ECG analog signal to obtain fetal ECG components;

In one possible embodiment, when the maternal ECG analog signal belonging to the current window signal is composed of the maternal ECG signal template of the current window signal, the fetal ECG component calculation module uses the maternal ECG signal of the current window signal The templates are stitched together to form a maternal ECG analog signal consistent with the time length of the next window signal;

In a certain embodiment, a maternal ECG signal template is added before the maternal ECG analog signal;

In a certain embodiment, a maternal ECG signal template located at the end of the maternal ECG analog signal is completed;

In a certain embodiment, the fetal ECG component calculation module is also specifically used to align the R peak of the first maternal ECG signal of the current window signal with the position of the R peak of the maternal ECG signal with the maternal ECG simulation The position of the second R-peak of the signal.

In one possible embodiment, the fetal ECG signal extraction system further includes a preprocessing module, which is used to preprocess the abdominal signal of the pregnant woman after collecting the abdominal signal of the pregnant woman, or divide the abdominal signal of the pregnant woman to obtain the current window signal, Preprocessing the abdominal signal of pregnant women;

In a certain embodiment, the method of preprocessing includes processing baseline, smoothing signal curve and/or filtering clutter;

In one embodiment, wavelet transform is used for filtering and smoothing.

It should be understood that before and after in the above embodiments refer to a time relationship.

The above-mentioned embodiment is only a preferred embodiment of the present invention, and cannot be used to limit the protection scope of the present invention. Any insubstantial changes and substitutions made by those skilled in the art on the basis of the present invention belong to the scope of the present invention. Scope of protection claimed.

Claims (13)

1. A fetal electrocardiosignal extraction device, comprising:

the acquisition module is used for acquiring abdomen signals of pregnant women;

the window signal segmentation module is used for segmenting the pregnant woman abdomen signal to obtain a current window signal, and searching the R wave position of the maternal electrocardiosignal in the pregnant woman abdomen signal through the current window signal;

the signal segment segmentation module is used for segmenting the current window signal according to the R wave position of the maternal electrocardiosignal to obtain a plurality of signal segments;

the parent electrocardiosignal template construction module is used for reconstructing a parent electrocardiosignal template belonging to the current window signal according to the signal segment in the current window signal;

the fetal electrocardio component calculation module is used for calculating and obtaining the fetal electrocardio component in the next window signal according to the maternal electrocardio signal template of the current window signal,

the window signal segmentation module extracts peaks in the window signal through a first-order differential sequence of the current window signal, takes peak amplitude values as clustering variables, and extracts the positions of R wave peaks of the maternal electrocardiosignals by using a clustering algorithm;

the window signal segmentation module determines the number of separated clusters, selects the cluster center of each cluster, calculates the Euclidean distance from the R wave peak to each cluster center, acquires the cluster center corresponding to each R wave peak according to the Euclidean distance from each R wave peak to each cluster center, calculates the sum of the Euclidean distances from the R wave peak to the corresponding cluster center, and selects another new cluster center;

comparing R wave peak to the Euclidean distance sum of the new corresponding cluster center and the old corresponding cluster center, and selecting a smaller one;

finally, the cluster center with the closest Euclidean distance sum from the R wave peak to the corresponding cluster center is obtained, and then the cluster center with the largest x value and the R wave peak corresponding to the cluster center are selected and defined as the R wave peak of the maternal electrocardiosignal.

2. The fetal electrocardiosignal extraction device of claim 1 wherein the first m present window signals of the present window signals obtained by dividing the pregnant woman abdomen signals are first window signals, and the m following present window signals are second window signals; the first window signal length is T1, the second window signal length is T2, and T1> T2.

3. The fetal heart signal extraction device of claim 1, further comprising a screening module for screening the signal segments according to a preset threshold;

the screening module selects the T meeting the requirement of b less than or equal to T _RR A signal segment of "b" or "a", wherein "b" or "T" will be satisfied _RR The n time periods less than or equal to a' are arranged according to the time length, the arranged signal periods are Xi, i=1-n, and T _RR A is X (n- ([ n/10)]Rounding)) and b is X ([ n/10)]Rounding) the length of time.

4. The fetal electrocardiosignal extraction device of claim 1 wherein the maternal electrocardiosignal template construction module obtains a standard time length according to the time length of each signal segment in the current window signal; extending or compressing the time length of each signal segment to a standard time length; obtaining a parent electrocardiosignal template according to the extended or compressed signal section;

and/or the maternal electrocardiosignal template construction module averages the extended or compressed signal segments at the same time point and then forms the average value into the maternal electrocardiosignal template.

5. The fetal heart signal extraction device of claim 4 wherein the maternal heart signal template construction module arranges the signal segments according to the length of time, and selects the signal segment that is located in the middle as the standard length of time.

6. The fetal heart rate signal extraction apparatus of claim 5 wherein the mother is configured to, when the number of signal segments is an odd numberThe body electrocardiosignal template construction module is ordered as the firstThe time length of each signal segment is the standard time length; when the number of the signal segments is even, the parent electrocardiosignal template construction module takes the sequence of +.>The time length of each signal segment is the standard time length; where n is the number of signal segments in the current window signal.

7. The fetal electrocardiosignal extraction device of claim 1 wherein the fetal electrocardiosignal component calculation module forms a maternal electrocardiosignal analog signal belonging to the current window signal through a maternal electrocardiosignal template of the current window signal; subtracting the maternal electrocardio analog signal of the current window signal from the next window signal to obtain the fetal electrocardio component of the next window signal.

8. The apparatus according to claim 7, wherein the fetal electrocardiosignal calculation module forms a maternal electrocardiosignal analog signal consistent with the time length of the next window signal by stitching the maternal electrocardiosignal template of the current window signal when the maternal electrocardiosignal template of the current window signal forms the maternal electrocardiosignal analog signal belonging to the current window signal.

9. The fetal heart signal extraction device of claim 8, wherein the fetal heart signal component calculation module adds a maternal heart signal template to the maternal heart signal analog signal.

10. A fetal heart signal extraction apparatus as claimed in claim 9 wherein the fetal heart signal component calculation module complements a maternal heart signal template located at the end of the maternal heart signal analog signal.

11. The fetal heart signal extraction device of claim 10 wherein the fetal heart signal component calculation module aligns the R-peak of the first maternal heart signal with the R-peak of the second maternal heart signal of the current window signal for which the R-peak of the maternal heart signal has been found.

12. The fetal heart signal extraction device of claim 1, further comprising:

the preprocessing module is used for preprocessing the pregnant woman abdomen signals after collecting the pregnant woman abdomen signals, or preprocessing the pregnant woman abdomen signals after dividing the pregnant woman abdomen signals to obtain window signals;

the method of preprocessing includes processing a baseline, smoothing a signal curve, and/or filtering clutter;

and/or filtering smoothing processing is performed by wavelet transformation.

13. A fetal electrocardiosignal extraction method, which is characterized by comprising the following steps:

collecting abdomen signals of pregnant women;

dividing the abdomen signal of the pregnant woman to obtain a current window signal, and searching the R wave position of the mother electrocardiosignal;

segmenting the current window signal according to the R wave position of the parent electrocardiosignal to obtain a plurality of signal segments;

reconstructing a parent electrocardiosignal template belonging to the current window signal according to the signal segment in the current window signal;

calculating to obtain fetal electrocardio components in the next window signal according to a maternal electrocardio signal template of the current window signal;

when the R wave position of the maternal electrocardiosignal is found, the peak in the window signal is extracted through the first-order differential sequence of the current window signal, the peak amplitude is used as a clustering variable, and the position of the R wave peak of the maternal electrocardiosignal is extracted by using a clustering algorithm;

for determining the number of separate clusters, selecting a cluster center for each cluster; calculating the Euclidean distance from the R wave peak to each cluster center;

acquiring a cluster center corresponding to each R wave peak according to the Euclidean distance from each R wave peak to each cluster center;

calculating the sum of Euclidean distances from the R wave peak to the corresponding clustering centers;

selecting another new clustering center;

comparing R wave peak to the Euclidean distance sum of the new corresponding cluster center and the old corresponding cluster center, and selecting a smaller one;

finally, the cluster center from the peak to the nearest cluster center from the sum is obtained, and then the cluster center with the maximum x value and the corresponding R wave peak are selected to be defined as the R wave peak of the maternal electrocardiosignal.